Face Generation

In this project, you'll use generative adversarial networks to generate new images of faces.

Get the Data

You'll be using two datasets in this project:

  • MNIST
  • CelebA

Since the celebA dataset is complex and you're doing GANs in a project for the first time, we want you to test your neural network on MNIST before CelebA. Running the GANs on MNIST will allow you to see how well your model trains sooner.

If you're using FloydHub, set data_dir to "/input" and use the FloydHub data ID "R5KrjnANiKVhLWAkpXhNBe".

In [2]:
data_dir = './data'

# FloydHub - Use with data ID "R5KrjnANiKVhLWAkpXhNBe"
#data_dir = '/input'


"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper

helper.download_extract('mnist', data_dir)
helper.download_extract('celeba', data_dir)
Found mnist Data
Found celeba Data

Explore the Data

MNIST

As you're aware, the MNIST dataset contains images of handwritten digits. You can view the first number of examples by changing show_n_images.

In [3]:
show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
%matplotlib inline
import os
from glob import glob
from matplotlib import pyplot
!pip install -U matplotlib==2.0.2

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')
Requirement already up-to-date: matplotlib==2.0.2 in /opt/conda/lib/python3.6/site-packages
Requirement already up-to-date: six>=1.10 in /opt/conda/lib/python3.6/site-packages (from matplotlib==2.0.2)
Requirement already up-to-date: python-dateutil in /opt/conda/lib/python3.6/site-packages (from matplotlib==2.0.2)
Requirement already up-to-date: pytz in /opt/conda/lib/python3.6/site-packages (from matplotlib==2.0.2)
Requirement already up-to-date: pyparsing!=2.0.0,!=2.0.4,!=2.1.2,!=2.1.6,>=1.5.6 in /opt/conda/lib/python3.6/site-packages (from matplotlib==2.0.2)
Requirement already up-to-date: numpy>=1.7.1 in /opt/conda/lib/python3.6/site-packages (from matplotlib==2.0.2)
Requirement already up-to-date: cycler>=0.10 in /opt/conda/lib/python3.6/site-packages/cycler-0.10.0-py3.6.egg (from matplotlib==2.0.2)
You are using pip version 9.0.1, however version 10.0.1 is available.
You should consider upgrading via the 'pip install --upgrade pip' command.
Out[3]:
<matplotlib.image.AxesImage at 0x7fbc480e7e48>

CelebA

The CelebFaces Attributes Dataset (CelebA) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing show_n_images.

In [4]:
show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))
Out[4]:
<matplotlib.image.AxesImage at 0x7fbc422d37b8>

Preprocess the Data

Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.

The MNIST images are black and white images with a single color channel while the CelebA images have 3 color channels (RGB color channel).

Build the Neural Network

You'll build the components necessary to build a GANs by implementing the following functions below:

  • model_inputs
  • discriminator
  • generator
  • model_loss
  • model_opt
  • train

Check the Version of TensorFlow and Access to GPU

This will check to make sure you have the correct version of TensorFlow and access to a GPU

In [5]:
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer.  You are using {}'.format(tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))
TensorFlow Version: 1.3.0
Default GPU Device: /gpu:0

Input

Implement the model_inputs function to create TF Placeholders for the Neural Network. It should create the following placeholders:

  • Real input images placeholder with rank 4 using image_width, image_height, and image_channels.
  • Z input placeholder with rank 2 using z_dim.
  • Learning rate placeholder with rank 0.

Return the placeholders in the following the tuple (tensor of real input images, tensor of z data)

In [6]:
import problem_unittests as tests

def model_inputs(image_width, image_height, image_channels, z_dim):
    """
    Create the model inputs
    :param image_width: The input image width
    :param image_height: The input image height
    :param image_channels: The number of image channels
    :param z_dim: The dimension of Z
    :return: Tuple of (tensor of real input images, tensor of z data, learning rate)
    """
    # TODO: Implement Function
    input_real = tf.placeholder(tf.float32, (None, image_width, image_height, image_channels), name='input_real')
    input_z = tf.placeholder(tf.float32, (None, z_dim), name='input_z')
    input_lr = tf.placeholder(tf.float32, (), name='input_lr')
    
    return (input_real, input_z, input_lr)


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_inputs(model_inputs)
Tests Passed

Discriminator

Implement discriminator to create a discriminator neural network that discriminates on images. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "discriminator" to allow the variables to be reused. The function should return a tuple of (tensor output of the discriminator, tensor logits of the discriminator).

In [8]:
def discriminator(images, reuse=False, alpha=0.05, is_training=True):
    """
    Create the discriminator network
    :param images: Tensor of input image(s)
    :param reuse: Boolean if the weights should be reused
    :return: Tuple of (tensor output of the discriminator, tensor logits of the discriminator)
    """
    # TODO: Implement Function
    
    with tf.variable_scope('discriminator', reuse=reuse):
        
        x1 = tf.layers.conv2d(inputs=images, strides=2, filters=128, kernel_size=3, padding='same')
        x1 = tf.maximum(alpha * x1, x1)
        x1 = tf.layers.dropout(x1)

        x2 = tf.layers.conv2d(inputs=x1, strides=2, filters=256, kernel_size=3, padding='same')
        x2 = tf.layers.batch_normalization(x2, training=is_training)
        x2 = tf.maximum(alpha * x2, x2)
        x2 = tf.layers.dropout(x2)

        x3 = tf.layers.conv2d(inputs=x2, strides=2, filters=512, kernel_size=3, padding='valid')
        x3 = tf.layers.batch_normalization(x3, training=is_training)
        x3 = tf.maximum(alpha * x3, x3)

        flat = tf.reshape(x3, (-1, 4*4*512))
        logits = tf.layers.dense(inputs=flat, units=1, activation=None)
        out = tf.sigmoid(logits)

    return (out, logits)


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_discriminator(discriminator, tf)
Tests Passed

Generator

Implement generator to generate an image using z. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "generator" to allow the variables to be reused. The function should return the generated 28 x 28 x out_channel_dim images.

In [9]:
def generator(z, out_channel_dim, is_train=True, alpha=0.05, reuse=False):
    """
    Create the generator network
    :param z: Input z
    :param out_channel_dim: The number of channels in the output image
    :param is_train: Boolean if generator is being used for training
    :return: The tensor output of the generator
    """
    # TODO: Implement Function
    with tf.variable_scope('generator', reuse=not is_train):
        #first layer - dense
        x1 = tf.layers.dense(inputs=z, units=3*3*512)
        
        x1 = tf.reshape(x1, (-1, 3, 3, 512))
        x1 = tf.layers.batch_normalization(inputs=x1, training=is_train)
        x1 = tf.maximum(alpha * x1, x1)
        x1 = tf.layers.dropout(x1)
        
        #second layer - convolutional
        x2 = tf.layers.conv2d_transpose(inputs=x1, strides=2, kernel_size=3, filters=256, padding='valid')
        x2 = tf.layers.batch_normalization(inputs=x2, training=is_train)
        x2 = tf.maximum(alpha * x2, x2)
        x2 = tf.layers.dropout(x2)
        
        #third layers - convolutional
        x3 = tf.layers.conv2d_transpose(inputs=x2, strides=2, kernel_size=3, filters=128, padding='same')
        x3 = tf.layers.batch_normalization(inputs=x3, training=is_train)
        x3 = tf.maximum(alpha * x3, x3)
        
        #third layer - convolutional
        x3 = tf.layers.conv2d_transpose(inputs=x3, strides=2, kernel_size=3, filters=out_channel_dim, padding='same')
        logits = tf.tanh(x3)
        
    return logits


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_generator(generator, tf)
Tests Passed

Loss

Implement model_loss to build the GANs for training and calculate the loss. The function should return a tuple of (discriminator loss, generator loss). Use the following functions you implemented:

  • discriminator(images, reuse=False)
  • generator(z, out_channel_dim, is_train=True)
In [10]:
def model_loss(input_real, input_z, out_channel_dim):
    """
    Get the loss for the discriminator and generator
    :param input_real: Images from the real dataset
    :param input_z: Z input
    :param out_channel_dim: The number of channels in the output image
    :return: A tuple of (discriminator loss, generator loss)
    """
    # TODO: Implement Function
    g_logits = generator(input_z, out_channel_dim, is_train=True)
    d_out_real, d_logits_real = discriminator(input_real, reuse=False)
    d_out_fake, d_logits_fake = discriminator(g_logits, reuse=True)
    
    d_loss_real = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_real, labels=tf.ones_like(d_out_real)))
    d_loss_fake = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.zeros_like(d_out_fake)))
    g_loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.ones_like(d_out_fake)))
    d_loss = d_loss_real + d_loss_fake
    
    return (d_loss, g_loss)


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_loss(model_loss)
Tests Passed

Optimization

Implement model_opt to create the optimization operations for the GANs. Use tf.trainable_variables to get all the trainable variables. Filter the variables with names that are in the discriminator and generator scope names. The function should return a tuple of (discriminator training operation, generator training operation).

In [11]:
def model_opt(d_loss, g_loss, learning_rate, beta1):
    """
    Get optimization operations
    :param d_loss: Discriminator loss Tensor
    :param g_loss: Generator loss Tensor
    :param learning_rate: Learning Rate Placeholder
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :return: A tuple of (discriminator training operation, generator training operation)
    """
    # TODO: Implement Function
    t_vars = tf.trainable_variables()
    g_vars = [var for var in t_vars if var.name.startswith('generator')]
    d_vars = [var for var in t_vars if var.name.startswith('discriminator')]
    
    with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):
        d_opt = tf.train.AdamOptimizer(learning_rate, beta1).minimize(d_loss, var_list=d_vars)
        g_opt = tf.train.AdamOptimizer(learning_rate, beta1).minimize(g_loss, var_list=g_vars)
    
    return (d_opt, g_opt)


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_opt(model_opt, tf)
Tests Passed

Neural Network Training

Show Output

Use this function to show the current output of the generator during training. It will help you determine how well the GANs is training.

In [12]:
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import numpy as np

def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
    """
    Show example output for the generator
    :param sess: TensorFlow session
    :param n_images: Number of Images to display
    :param input_z: Input Z Tensor
    :param out_channel_dim: The number of channels in the output image
    :param image_mode: The mode to use for images ("RGB" or "L")
    """
    cmap = None if image_mode == 'RGB' else 'gray'
    z_dim = input_z.get_shape().as_list()[-1]
    example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])

    samples = sess.run(
        generator(input_z, out_channel_dim, False),
        feed_dict={input_z: example_z})

    images_grid = helper.images_square_grid(samples, image_mode)
    pyplot.imshow(images_grid, cmap=cmap)
    pyplot.show()

Train

Implement train to build and train the GANs. Use the following functions you implemented:

  • model_inputs(image_width, image_height, image_channels, z_dim)
  • model_loss(input_real, input_z, out_channel_dim)
  • model_opt(d_loss, g_loss, learning_rate, beta1)

Use the show_generator_output to show generator output while you train. Running show_generator_output for every batch will drastically increase training time and increase the size of the notebook. It's recommended to print the generator output every 100 batches.

In [13]:
import pickle as pkl

def train(epoch_count, batch_size, z_dim, learning_rate, beta1, get_batches, data_shape, data_image_mode):
    """
    Train the GAN
    :param epoch_count: Number of epochs
    :param batch_size: Batch Size
    :param z_dim: Z dimension
    :param learning_rate: Learning Rate
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :param get_batches: Function to get batches
    :param data_shape: Shape of the data
    :param data_image_mode: The image mode to use for images ("RGB" or "L")
    """
    # TODO: Build Model
    input_real, input_z, input_lr = model_inputs(data_shape[1], data_shape[2], data_shape[3], z_dim)
    d_loss, g_loss = model_loss(input_real, input_z, data_shape[3])
    d_opt, g_opt = model_opt(d_loss, g_loss, learning_rate, beta1)
    
    # TODO: Train Model
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        for epoch_i in range(epoch_count):
            count = 0
            for batch_images in get_batches(batch_size):
                batch_images = batch_images * 2
                count = count + 1
                batch_z = np.random.uniform(-1, 1, (batch_size, z_dim))
                
                _ = sess.run(d_opt, feed_dict={input_real:batch_images, input_z:batch_z})
                _ = sess.run(g_opt, feed_dict={input_z:batch_z, input_real:batch_images})
                
                #print after every 25 iterations
                if count % 400 == 0:
                    train_loss_d = d_loss.eval({input_real:batch_images, input_z:batch_z})
                    train_loss_g = g_loss.eval({input_z:batch_z})
                    
                    print('Batch/Epoch {}/{}'.format(count, epoch_i), 
                          '\t Discriminator Loss: {}'.format(train_loss_d), 
                          '\t Generator Loss: {}'.format(train_loss_g))
                
                if count % 100 == 0:
                    _ = show_generator_output(sess, 20, input_z, data_shape[3], data_image_mode)
                

MNIST

Test your GANs architecture on MNIST. After 2 epochs, the GANs should be able to generate images that look like handwritten digits. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

In [14]:
batch_size = 32
z_dim = 70
learning_rate = 0.003
beta1 = 0.75


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 2

mnist_dataset = helper.Dataset('mnist', glob(os.path.join(data_dir, 'mnist/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, mnist_dataset.get_batches,
          mnist_dataset.shape, mnist_dataset.image_mode)
Batch/Epoch 400/0 	 Discriminator Loss: 0.8120257258415222 	 Generator Loss: 1.4694349765777588
Batch/Epoch 800/0 	 Discriminator Loss: 0.9216799139976501 	 Generator Loss: 0.9164689183235168
Batch/Epoch 1200/0 	 Discriminator Loss: 0.6626117825508118 	 Generator Loss: 0.9549350142478943
Batch/Epoch 1600/0 	 Discriminator Loss: 0.7976991534233093 	 Generator Loss: 1.158623218536377
Batch/Epoch 400/1 	 Discriminator Loss: 0.28798481822013855 	 Generator Loss: 1.7775726318359375
Batch/Epoch 800/1 	 Discriminator Loss: 0.2492416799068451 	 Generator Loss: 2.590205669403076
Batch/Epoch 1200/1 	 Discriminator Loss: 0.2759729325771332 	 Generator Loss: 2.5478591918945312
Batch/Epoch 1600/1 	 Discriminator Loss: 0.2352239042520523 	 Generator Loss: 2.844866991043091

CelebA

Run your GANs on CelebA. It will take around 20 minutes on the average GPU to run one epoch. You can run the whole epoch or stop when it starts to generate realistic faces.

In [16]:
batch_size = 32
z_dim = 110
learning_rate = 0.0003
beta1 = 0.75


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 1

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)
Batch/Epoch 400/0 	 Discriminator Loss: 0.2741240859031677 	 Generator Loss: 1.89737868309021
Batch/Epoch 800/0 	 Discriminator Loss: 0.22081258893013 	 Generator Loss: 2.459805727005005
Batch/Epoch 1200/0 	 Discriminator Loss: 0.06285504996776581 	 Generator Loss: 3.6464991569519043
Batch/Epoch 1600/0 	 Discriminator Loss: 0.41549152135849 	 Generator Loss: 1.4868733882904053
Batch/Epoch 2000/0 	 Discriminator Loss: 0.48971879482269287 	 Generator Loss: 1.8133580684661865
Batch/Epoch 2400/0 	 Discriminator Loss: 0.6190822720527649 	 Generator Loss: 1.1086210012435913
Batch/Epoch 2800/0 	 Discriminator Loss: 0.5058559775352478 	 Generator Loss: 1.3676120042800903
Batch/Epoch 3200/0 	 Discriminator Loss: 0.3827211260795593 	 Generator Loss: 1.5517381429672241
Batch/Epoch 3600/0 	 Discriminator Loss: 0.3037433624267578 	 Generator Loss: 2.3090426921844482
Batch/Epoch 4000/0 	 Discriminator Loss: 0.4292309880256653 	 Generator Loss: 2.0635457038879395
Batch/Epoch 4400/0 	 Discriminator Loss: 0.6836898922920227 	 Generator Loss: 1.2964506149291992
Batch/Epoch 4800/0 	 Discriminator Loss: 0.6657252311706543 	 Generator Loss: 1.3543744087219238
Batch/Epoch 5200/0 	 Discriminator Loss: 0.3617731034755707 	 Generator Loss: 1.5507302284240723
Batch/Epoch 5600/0 	 Discriminator Loss: 0.5055624842643738 	 Generator Loss: 1.3583705425262451
Batch/Epoch 6000/0 	 Discriminator Loss: 0.8755422830581665 	 Generator Loss: 1.0934016704559326

Submitting This Project

When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_face_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.